Carbonate rock electric pump well full-process multi-dimensional production analysis adjustment method and system

Abstract

The invention discloses a carbonate rock electric pump well full-process multi-dimensional production analysis adjustment method and system, and the method comprises the steps: obtaining electric pump well related data including geological data, oil reservoir data and the like, and forming an integrated electric pump well data frame; a mechanism diagnosis logic sub-model, a current and voltage diagnosis sub-model, a pump parameter diagnosis sub-model, a pressure yield diagnosis sub-model and a vibration diagnosis sub-model are electrically constructed, and the sub-models are fused into a comprehensive diagnosis model; inputting data of the electric pump well integrated database into a comprehensive diagnosis model, and judging whether the electric pump well breaks down or not; if the electric pump well breaks down, fault generation reason analysis is conducted according to the fault type of the electric pump well in combination with the recognition logic table, and follow-up measure optimization is conducted according to different fault reasons and the fault type of the electric pump well. According to the method, the fault type of the electric pump can be accurately identified in combination with geological reservoir related information, the fault generation reason of the electric pump can be accurately analyzed, and production optimization adjustment of an electric pump well can be further guided.

Description

Technical Field

[0001] The present invention relates to the technical field of electric pump well production management, and in particular to a full-process multi-dimensional production analysis and adjustment method and system for carbonate rock electric pump wells. Background Art

[0002] The artificial lifting process of electric submersible pumps is a key technology for the later development of carbonate oil fields. Due to its strong applicability, convenient maintenance and management, and high lifting efficiency, it is widely used in major oil fields at home and abroad. Since the electric pump runs at the bottom of the well for a long time, daily precise maintenance and management is conducive to improving the operating life of the electric pump, further reducing the number of electric pump overhauls, and reducing maintenance costs. Through accurate monitoring and detailed analysis of the real-time production data of the electric pump well, such as displacement, head, current, voltage and other key indicators, scientific methods are used to optimize production parameters to ensure that the electric pump well operates under the best working conditions. This not only helps to improve the liquid production efficiency of a single well, but also effectively reduces energy consumption, thereby improving the overall production efficiency and economic benefits of the oil field. On the basis of electric pump maintenance management, an effective production status analysis method for electric pump wells is established. Based on the results of electric pump fault analysis and production status judgment, the production dynamics of electric pump wells are guided to be efficiently adjusted, further liberating and improving the productivity of electric pump wells.

[0003] Previous production management methods for electric pump wells were limited to the analysis of electric pump data, without further combining geological and reservoir-related information. As a result, only the type of electric pump failure could be roughly obtained, making it difficult to accurately analyze the cause of the electric pump failure and further guide the production optimization and adjustment of the electric pump wells.

[0004] In view of this, this application is hereby filed. Summary of the invention

[0005] The purpose of the present invention is to provide a multi-dimensional production analysis and adjustment method and system for the whole process of carbonate rock electric pump wells. The present invention first establishes an integrated electric pump well data framework (geology-reservoir-wellbore-electric pump data framework), and uses the dynamic data in the framework to establish a mechanism diagnosis logic sub-model (i.e., the reaction logic of electric pump well production fault identification-fault cause analysis-adjustment measure suggestion); secondly, a neural network is used to construct a current and voltage diagnosis sub-model, a pump parameter diagnosis sub-model, a pressure and production diagnosis sub-model, and a vibration diagnosis sub-model; then, the above sub-models are also integrated into a comprehensive diagnosis model; finally, the electric pump well fault diagnosis is performed based on the comprehensive diagnosis model. The present invention can accurately identify the type of electric pump fault, accurately analyze the cause of the electric pump fault, and further guide the production optimization and adjustment of the electric pump well in combination with geological reservoir related information.

[0006] The present invention is achieved through the following technical solutions:

[0007] In a first aspect, the present invention provides a full-process multi-dimensional production analysis and adjustment method for a carbonate rock electric pump well, the method comprising:

[0008] Acquire the relevant data of the electric pump well, form an integrated electric pump well data framework and store it in an integrated electric pump well database; the relevant data of the electric pump well includes geological data, reservoir data, wellbore data and electric pump data;

[0009] Based on the dynamic data in the integrated electric pump well data framework, a mechanism diagnosis logic sub-model is constructed; the mechanism diagnosis logic sub-model is an identification logic table composed of preset electric pump well fault types, fault causes and adjustment measures;

[0010] Based on the integrated electric pump well data framework, neural networks are used to construct current and voltage diagnosis sub-models, pump parameter diagnosis sub-models, pressure and production diagnosis sub-models, and vibration diagnosis sub-models.

[0011] Using deep learning neural network, the mechanism diagnosis logic sub-model, current and voltage diagnosis sub-model, pump parameter diagnosis sub-model, pressure and output diagnosis sub-model and vibration diagnosis sub-model are integrated into a comprehensive diagnosis model.

[0012] The data of the electric pump well integrated database is input into the comprehensive diagnosis model to determine whether the electric pump well has a fault; if the electric pump well has a fault, the cause of the fault is analyzed based on the fault type of the electric pump well and combined with the identification logic table, and subsequent measures are optimized based on different fault causes and fault types of the electric pump well.

[0013] Furthermore, the method further comprises:

[0014] If the electric pump well has not failed, the electric pump well that has not failed will be scored through the electric pump well scoring and evaluation system, and attention will be paid to the operating status of the electric pump well with a low score.

[0015] Furthermore, the electric pump wells without faults are scored through the electric pump well operation scoring evaluation system, and the operation status of the electric pump wells with low scores is paid attention to, including:

[0016] The running state of the electric pump well without any fault is scored by the electric pump well operation scoring evaluation system to obtain the electric pump well score;

[0017] For electric pump wells with low electric pump well scores, the number of oilfield electric pump failures is determined through the electric pump well fault evaluation method, and the pump inspection cycles of different types of electric pumps are statistically obtained as the fault evaluation results;

[0018] Based on the fault evaluation results and the static data in the electric pump well related data, the operation effect of the electric pump is evaluated, and the electric pump wells that are of particular concern are obtained;

[0019] Among them, static data include production layer, fluid density, well type, injection-production ratio, pump hanging depth, asphaltene and gas-liquid ratio.

[0020] Furthermore, the operation effect of the electric pump is evaluated, including:

[0021] The stable operation degree of the electric pump is evaluated as follows: the pump inspection cycle under the screening conditions / the average pump inspection cycle of the entire oil field × 100%. The failure rate of the electric pump is as follows: the number of failures of the electric pump under the screening conditions / the total number of failures of the entire oil field × 100%.

[0022] The failure rate of the electric pump was evaluated as follows: the number of electric pump failures under the screening conditions / the total number of failures in the entire oil field × 100%.

[0023] Furthermore, the fault types of electric pump wells include electric pump failure, fluid change failure, unit and pipe string failure;

[0024] Among them, electric pump failures include pump blockage, underload, overload, load fluctuation, overcurrent and undervoltage failures;

[0025] Fluid changes include fluid impurities and gas content increase faults;

[0026] Unit and tubing failures include unit aging, undercurrent protection failure, overcurrent protection failure, pump disconnection, downhole tubing leakage and impeller wear failure.

[0027] Furthermore, the steps of constructing the current and voltage diagnosis sub-model include:

[0028] Batch read the electric pump operation data from the electric pump well integrated database, and perform data preprocessing to obtain time column and current column data;

[0029] Based on the time column and current column data, coordinate transformation is performed and batch drawing is performed into current card diagrams of standard pixel size;

[0030] The coordinate transformation is to plot the collected current train data that changes with time into radial coordinates.

[0031] Furthermore, the transformation expression of coordinate transformation is:

[0032]

[0033] In the formula, x is the horizontal axis value; y is the vertical axis value; t is the time of the day, converted to the hour system; A is the current.

[0034] In a second aspect, the present invention provides a full-process multi-dimensional production analysis and adjustment system for carbonate rock electric pump wells, the system comprising:

[0035] An acquisition unit is used to acquire electric pump well related data, form an integrated electric pump well data framework and store it in an electric pump well integrated database; the electric pump well related data includes geological data, reservoir data, wellbore data and electric pump data;

[0036] The first sub-model unit is used to construct a mechanism diagnosis logic sub-model based on the dynamic data in the integrated electric pump well data framework; the mechanism diagnosis logic sub-model is an identification logic table composed of preset electric pump well fault types, fault causes and adjustment measures;

[0037] The second sub-model unit is used to construct a current and voltage diagnosis sub-model, a pump parameter diagnosis sub-model, a pressure and production diagnosis sub-model and a vibration diagnosis sub-model respectively based on the integrated electric pump well data framework by using a neural network;

[0038] A model fusion unit, used to fuse the mechanism diagnosis logic sub-model, the current and voltage diagnosis sub-model, the pump parameter diagnosis sub-model, the pressure and output diagnosis sub-model and the vibration diagnosis sub-model into a comprehensive diagnosis model by using a deep learning neural network;

[0039] The fault judgment unit is used to input the data of the electric pump well integrated database into the comprehensive diagnosis model to determine whether the electric pump well has a fault; if a fault occurs in the electric pump well, the cause of the fault is analyzed based on the fault type of the electric pump well and combined with the identification logic table, and subsequent measures are optimized based on different fault causes and fault types of the electric pump well.

[0040] Furthermore, the system also includes:

[0041] The electric pump operation evaluation unit is used to score the electric pump wells that have not failed through the electric pump well scoring evaluation system if the electric pump well has not failed, and pay attention to the operation status of the electric pump wells with low scores, including:

[0042] The running state of the electric pump well without any fault is scored by the electric pump well operation scoring evaluation system to obtain the electric pump well score;

[0043] For electric pump wells with low electric pump well scores, the number of oilfield electric pump failures is determined through the electric pump well fault evaluation method, and the pump inspection cycles of different types of electric pumps are statistically obtained as the fault evaluation results;

[0044] Based on the fault evaluation results and the static data in the electric pump well related data, the operation effect of the electric pump is evaluated, and the electric pump wells that are of particular concern are obtained;

[0045] Among them, static data include production layer, fluid density, well type, injection-production ratio, pump hanging depth, asphaltene and gas-liquid ratio.

[0046] In a third aspect, the present invention further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned full-process multi-dimensional production analysis and adjustment method for carbonate electric pump wells.

[0047] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0048] 1. The present invention provides a multi-dimensional production analysis and adjustment method and system for the entire process of carbonate rock electric pump wells. First, an integrated electric pump well data framework (geology-reservoir-wellbore-electric pump data framework) is established. The dynamic data in the framework is used to establish a mechanism diagnosis logic sub-model (i.e., the reaction logic of electric pump well production fault identification-fault cause analysis-adjustment measure suggestion); secondly, a neural network is used to respectively construct a current and voltage diagnosis sub-model, a pump parameter diagnosis sub-model, a pressure and production diagnosis sub-model, and a vibration diagnosis sub-model; then, the above sub-models are also integrated into a comprehensive diagnosis model; finally, the electric pump well fault diagnosis is performed based on the comprehensive diagnosis model. The present invention can accurately identify the type of electric pump fault, accurately analyze the cause of the electric pump fault, and further guide the production optimization and adjustment of the electric pump well in combination with geological reservoir-related information.

[0049] 2. The present invention provides a multi-dimensional production analysis and adjustment method and system for the entire process of carbonate rock electric pump wells. For electric pump wells that have not experienced any failure, the electric pump operation scoring evaluation system is used to score the operation status of the electric pump. For wells with lower scores, it is necessary to focus on the operation of the electric pump, which is very likely to have an electric pump failure. Through the electric pump well failure evaluation method, the number of oilfield electric pump failures can be determined, and the pump inspection cycles of different types of electric pumps can be obtained by statistics. The failure evaluation results are further combined with static data to score and evaluate the operation effect of the electric pump. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0051] Figure 1 This is a flow chart of a full-process multi-dimensional production analysis and adjustment method for carbonate rock electric pump wells according to Example 1 of the present invention;

[0052] Figure 2 The flow chart of batch generation of current cards of the present invention;

[0053] Figure 3 Schematic diagram of the current card of the present invention;

[0054] Figure 4 Schematic diagram of the convolutional neural network (CNN) of the present invention;

[0055] Figure 5The convolutional neural network (CNN) picture recognition flow chart of the present invention;

[0056] Figure 6 Flowchart of establishing LSTM prediction model of the present invention;

[0057] Figure 7 Embodiment 2 of the present invention is a detailed flow chart of a multi-dimensional production analysis and adjustment method for a carbonate rock electric pump well in the whole process;

[0058] Figure 8 This is a structural block diagram of a full-process multi-dimensional production analysis and adjustment system for carbonate rock electric pump wells according to the present invention. DETAILED DESCRIPTION

[0059] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0060] Previous production management methods for electric pump wells were limited to the analysis of electric pump data, without further combining geological and reservoir-related information. As a result, only the type of electric pump failure could be roughly obtained, making it difficult to accurately analyze the cause of the electric pump failure and further guide the production optimization and adjustment of the electric pump wells.

[0061] The present invention designs a multi-dimensional production analysis and adjustment method and system for the whole process of carbonate rock electric pump wells. The present invention first establishes an integrated electric pump well data framework (geology-reservoir-wellbore-electric pump data framework), uses the dynamic data in the framework, establishes a mechanism diagnosis logic sub-model (i.e., the reaction logic of electric pump well production fault identification-fault cause analysis-adjustment measure suggestion), and automatically imports production data through computer programming, which can not only quickly identify electric pump faults, but also quickly form measures adjustment suggestions; secondly, neural networks are used to construct current and voltage diagnosis sub-models, pump parameter diagnosis sub-models, pressure and production diagnosis sub-models, and vibration diagnosis sub-models; among them, considering that current data is an important parameter for electric pump well fault identification, a data-driven current card automatic generation and rapid identification method is established based on a convolutional neural network (CNN), which is used as an important basis for electric pump fault judgment. Then, the above sub-models are also integrated into a comprehensive diagnosis model; finally, the electric pump well fault diagnosis is performed based on the comprehensive diagnosis model. The dynamic data analysis results are combined with static data to further build an electric pump operation evaluation system to summarize, evaluate and analyze the operation status of the electric pump and the production status of the electric pump well.

[0062] The present invention can combine geological reservoir-related information to accurately identify the type of electric pump failure, accurately analyze the cause of the electric pump failure, and further guide the production optimization and adjustment of the electric pump well.

[0063] Example 1

[0064] like Figure 1 As shown, the present invention provides a full-process multi-dimensional production analysis and adjustment method for carbonate rock electric pump wells, the method comprising:

[0065] Step 1, obtaining electric pump well related data, forming an integrated electric pump well data framework and storing it in an electric pump well integrated database; the electric pump well related data includes geological data, reservoir data, wellbore data and electric pump data;

[0066] Specifically, to accurately analyze the production status of the electric pump well, it is necessary to fully grasp the relevant information of the electric pump well from the aspects of geology-reservoir-wellbore-electric pump, give measures and suggestions based on the production and operation of the electric pump, and further give process measures based on the geological reservoir information and wellbore conditions.

[0067] Geological data include geological types, fracture and fault development, etc.; reservoir data include production layer, average permeability of production layer, average porosity of production layer and historical production dynamic data, etc.; wellbore data include wellbore structure, well type and fluid properties, etc.; electric pump data include pump hanging depth, electric pump operation history data and electric pump failure history data, etc. The information included in the established data framework is shown in Table 1. The integrated electric pump well data framework includes both structured data such as production pressure and unstructured data such as operation logs, as well as single-day production data and hourly data, and is a multi-dimensional integrated database.

[0068] Table 1 Data types of integrated electric pump well data frame

[0069]

[0070]

[0071] Step 2: construct a mechanism diagnosis logic sub-model based on the dynamic data in the integrated electric pump well data framework; the mechanism diagnosis logic sub-model is an identification logic table composed of preset electric pump well fault types, fault causes and adjustment measures;

[0072] Specifically, dynamic data and some static data are extracted from the integrated electric pump well data framework for use in electric pump fault analysis and to guide the implementation of adjustment measures. The identification logic tables consisting of electric pump well fault types, fault causes and adjustment measures are shown in Tables 2 to 7.

[0073] In this embodiment, the types of electric pump well failures include electric pump failure, fluid change failure, unit and tubing failure; among them, electric pump failures include pump blockage, underload, overload, load fluctuation, overcurrent and underpressure failures; fluid changes include fluid impurities and gas content increase failures; unit and tubing failures include unit aging, undercurrent protection failure, overcurrent protection failure, pump peripheral disconnection, downhole tubing leakage and impeller wear failure.

[0074] The parameter change rules involved in the diagnostic logic of the mechanism diagnosis logic sub-model. Different parameters have different sensitivities to fault changes, and have different speeds of increase and decrease and time windows of change. However, the slope within a certain period of time is specified as the basis for judging parameter changes:

[0075]

[0076] C>C th , abnormal parameter changes

[0077] C<C th , almost unchanged

[0078] Among them, P ara is the parameter, such as pressure, output, temperature, current, etc.; t is the time, s; C is the parameter change measurement value; C th is the parameter change threshold;

[0079] The working capacity of each electric pump in each well is different, and the fixed threshold cannot truly reflect the situation of each well. Therefore, the dynamic change threshold is formulated based on the parameter change characteristics before the historical fault occurs. ara_start is the starting point of parameter change, P ara_fault Parameter value corresponding to electric pump failure, t start_fault is the fault response time, we get:

[0080]

[0081] Table 2 Electric pump well pump blockage adjustment logic

[0082]

[0083] Table 3 Electric pump well underload adjustment logic

[0084]

[0085] Table 4 Electric pump well overload adjustment logic

[0086]

[0087] Table 5 Overcurrent and overpressure adjustment logic of electric pump well

[0088]

[0089] Table 6 Adjustment logic of fluid problems in electric pump wells

[0090]

[0091]

[0092] Table 7 Fault adjustment logic of electric pump well unit

[0093]

[0094] Step 3, based on the integrated electric pump well data framework, a neural network is used to construct a current and voltage diagnosis sub-model, a pump parameter diagnosis sub-model, a pressure and production diagnosis sub-model, and a vibration diagnosis sub-model;

[0095] The key parameters that affect the failure changes of electric pump wells include motor current, wellhead oil production, oil pressure, and casing pressure. The current data is collected in real time through the wellhead VSD, which usually changes every second, while the wellhead oil production, oil pressure, and casing pressure usually change every day. Different data dimensions need to be classified and modeled.

[0096] Specifically, this embodiment uses a convolutional neural network in the current and voltage diagnosis sub-model to perform fault identification on current data as follows:

[0097] The current data is hourly data / second data, which can be used for real-time tracking of electric pump failures. A data-driven current card automatic identification method is established based on the machine learning algorithm to automatically identify the fault type of the electric pump well through the current card.

[0098] In this embodiment, the steps of constructing the current and voltage diagnosis sub-model include:

[0099] Batch read the electric pump operation data from the electric pump well integrated database, and perform data preprocessing to obtain time column and current column data;

[0100] Based on the time column and current column data, coordinate transformation is performed and batch drawing is performed into current card diagrams of standard pixel size;

[0101] The coordinate transformation is to plot the collected current train data that changes with time into radial coordinates.

[0102] In specific implementation, the data collected on site are usually saved in an Excel file in a folder named after a single well. An Excel file usually includes many days of electric pump operation data. Since engineers sample at different times each time, the Excel files in each folder are different, and the electric pump operation parameters contained in the Excel files are different. Therefore, it is necessary to establish a current card data batch processing process to batch process the collected electric pump operation data and batch draw current cards.

[0103] The excel file includes eight columns: Date / time, pi, pd, T, Vx, Vz, Vt, and Ct. The present invention mainly uses the time and current columns. The data in the excel file is extracted at 24h intervals starting from midnight, and then drawn into a current card diagram of standard pixel size. Then it is saved in .jpg format and named as pound sign + date. The specific processing flow is as follows: Figure 2 shown.

[0104] The collected current series data that changes with time is plotted into radial coordinates. The conversion expression of coordinate conversion is:

[0105]

[0106] In the formula, x is the horizontal axis value; y is the vertical axis value; t is the time of the day, converted to the hour system; A is the current.

[0107] The current card diagram is as follows Figure 3 shown.

[0108] Specifically, a convolutional neural network (CNN) is a deep learning architecture that is particularly suitable for processing data with a grid structure, such as images (2D grids) and time series (1D grids). The key feature of CNN is its ability to automatically and efficiently learn spatial hierarchical feature representations from data.

[0109] CNN mainly includes: Convolutional Layer, Weight Sharing, Activation Function, Pooling Layer, Fully Connected Layer, Local Receptive Field, which helps the model capture local features.

[0110] The operating mode of CNN is as follows Figure 4 As shown in the figure, CNN can effectively learn the spatial hierarchical features of images through this architecture, from simple edges and textures to complex object parts and overall structures, and is suitable for various image processing tasks such as image classification, object detection, image segmentation, and dynamometer diagnosis of rod pumps.

[0111] The calculation process of the convolutional neural network CNN current card recognition method is as follows Figure 5 As shown, it mainly includes the following steps:

[0112] Step 1: Import data. Integrate the data automatically collected on the cloud into a data set;

[0113] Step 2: Data preprocessing. Evaluate whether to limit the image pixels based on the amount of data and computer size, binarize the image, normalize it, and perform one-hot processing on the classification labels;

[0114] Step 4: Initial data set. Classify and save the processed data of each well to form an initial data set;

[0115] Step 5: Divide the data set into training set, validation set and test set with a size ratio of 7:2:1;

[0116] Step 6: CNN model training. The training set enters CNN training, and the test set is used to correct and strengthen the algorithm itself. The test set does not participate in training. After the training is completed, the test set is used for verification. If the ideal effect (accuracy above 90%) is achieved, the model is output. Otherwise, the CNN model parameters are changed and retrained.

[0117] Specifically, this embodiment also uses the long short-term memory neural network in the pressure-production diagnosis sub-model to illustrate the production and pressure parameter fault identification, as follows:

[0118] Since the failure of the electric pump well corresponds to the changing trend of production and pressure within a certain period of time, which is data with time series, the LSTM (Long Short-Term Memory Neural Network) algorithm is used for diagnosis. The LSTM (Long Short-Term Memory Neural Network) algorithm is an advanced recurrent neural network (RNN) used to process sequence data. Its core is the ability to capture long-term dependencies.

[0119] The fault diagnosis model is mainly established using the daily changing oil pressure, casing pressure, and liquid production. The collected data are pre-processed, mainly including the following aspects:

[0120] 1. Data cleaning: remove or correct errors and inconsistencies in the data. Missing values ​​are replaced by the average value of two adjacent points. Abnormal values ​​mainly include: the well is not shut in but the production or pressure is 0, and the pressure and production suddenly change to a smaller or larger value but immediately return to normal. After the abnormal values ​​are deleted, they are replaced by the average value of two adjacent points.

[0121] 2. Feature scaling: Standardize the data to keep the scales of different features consistent, thereby speeding up training and improving model performance.

[0122] 3. Time series decomposition: When processing time series data, the consistency of the time series length needs to be taken into account.

[0123] The diagnostic calculation steps of the LSTM model are:

[0124] Step 1: Collect the production pressure change data of a single well. The data collected by the wellhead equipment is automatically uploaded to the cloud;

[0125] Step 2: Select training features.

[0126] Step 3: Data preprocessing. Process the data using data cleaning, feature scaling, time series decomposition, etc.

[0127] Step 4: Initial data set. Classify and save the processed data of each well to form an initial data set;

[0128] Step 5: Divide the data set. After each well is classified individually, the data is divided into a training set, a validation set, and a test set with a size ratio of 7:2:1;

[0129] Step 6: LSTM model training. The training set enters LSTM training, and the test set is used to correct and strengthen the algorithm itself. The test set does not participate in training. After the training is completed, the test set is used for verification. If the ideal effect (accuracy above 90%) is achieved, the model is output. Otherwise, the LSTM model parameters are changed and retrained.

[0130] Step 7: Repeat Step 1-6 until all the models required for the subsequent steps are obtained;

[0131] Fault diagnosis and prediction methods for electric pump wells Figure 6 shown.

[0132] Step 4, using a deep learning neural network, the mechanism diagnosis logic sub-model, the current and voltage diagnosis sub-model, the pump parameter diagnosis sub-model, the pressure and output diagnosis sub-model, and the vibration diagnosis sub-model are integrated into a comprehensive diagnosis model;

[0133] Step 5, input the data of the electric pump well integrated database into the comprehensive diagnosis model to determine whether the electric pump well has a fault; if the electric pump well has a fault, analyze the cause of the fault based on the fault type of the electric pump well and the identification logic table, and make optimization suggestions for subsequent measures based on different fault causes and fault types of the electric pump well.

[0134] Since there are differences in human understanding of parameter changes in mechanism analysis, and the fault diagnosis results of analysis models established for different parameters will also be different, the present invention establishes a deep learning neural network to couple the mechanism diagnosis logic and multiple data-driven models to comprehensively understand the model results.

[0135] Example 2

[0136] like Figure 7 As shown, the difference between this embodiment and embodiment 1 is that the method further includes:

[0137] Step 6: If the electric pump well has not failed, the electric pump wells that have not failed are scored using the electric pump well scoring and evaluation system, and attention is paid to the operating status of the electric pump wells with low scores.

[0138] Step 6 specifically includes:

[0139] The motion state of the electric pump well without any fault is scored by the electric pump well operation scoring evaluation system to obtain the electric pump well score, as shown in Table 8.

[0140] For electric pump wells with low electric pump well scores, the number of oilfield electric pump failures is determined through the electric pump well fault evaluation method, and the pump inspection cycles of different types of electric pumps are statistically obtained as the fault evaluation results;

[0141] Based on the fault evaluation results and the static data in the relevant data of the electric pump wells, the operating effect of the electric pump is evaluated and the electric pump wells that are of particular concern are obtained; among which, the static data include production layer, fluid density, well type, injection-production ratio, pump hanging depth, asphaltene and gas-liquid ratio.

[0142] Among them, the stable operation degree of the electric pump is evaluated as follows: the pump inspection cycle under the screening conditions / the average pump inspection cycle of the entire oil field×100%, and the failure rate of the electric pump is as follows: the number of failures of the electric pump under the screening conditions / the total number of failures of the entire oil field×100%;

[0143] The failure rate of the electric pump was evaluated as follows: the number of electric pump failures under the screening conditions / the total number of failures in the entire oil field × 100%.

[0144] Table 8 Electric Pump Evaluation Scoring Table

[0145]

[0146] Single evaluation:

[0147]

[0148] Comprehensive evaluation:

[0149]

[0150] In the formula, ω i is the weight; max_ss is the highest score of a single item (15 in this table); n i is the score of a single item; max_s is the total maximum score (105 in this table);

[0151] In step 6 of the present invention, wells that have not experienced any failure are scored using a scoring evaluation system, and the operating status of wells with lower scores is focused on.

[0152] Example 3

[0153] like Figure 8 As shown, the difference between this embodiment and embodiment 1 is that this embodiment provides a full-process multi-dimensional production analysis and adjustment system for carbonate rock electric pump wells, and the functions of the system correspond one-to-one to the full-process multi-dimensional production analysis and adjustment method for carbonate rock electric pump wells in embodiment 1; the system includes:

[0154] An acquisition unit is used to acquire electric pump well related data, form an integrated electric pump well data framework and store it in an electric pump well integrated database; the electric pump well related data includes geological data, reservoir data, wellbore data and electric pump data;

[0155] The first sub-model unit is used to construct a mechanism diagnosis logic sub-model based on the dynamic data in the integrated electric pump well data framework; the mechanism diagnosis logic sub-model is an identification logic table composed of preset electric pump well fault types, fault causes and adjustment measures;

[0156] The second sub-model unit is used to construct a current and voltage diagnosis sub-model, a pump parameter diagnosis sub-model, a pressure and production diagnosis sub-model and a vibration diagnosis sub-model respectively based on the integrated electric pump well data framework by using a neural network;

[0157] A model fusion unit, used to fuse the mechanism diagnosis logic sub-model, the current and voltage diagnosis sub-model, the pump parameter diagnosis sub-model, the pressure and output diagnosis sub-model and the vibration diagnosis sub-model into a comprehensive diagnosis model by using a deep learning neural network;

[0158] The fault judgment unit is used to input the data of the electric pump well integrated database into the comprehensive diagnosis model to determine whether the electric pump well has a fault; if a fault occurs in the electric pump well, the cause of the fault is analyzed based on the fault type of the electric pump well and combined with the identification logic table, and subsequent measures are optimized based on different fault causes and fault types of the electric pump well.

[0159] As a further implementation, the system also includes:

[0160] The electric pump operation evaluation unit is used to score the electric pump wells that have not failed through the electric pump well scoring evaluation system if the electric pump well has not failed, and pay attention to the operation status of the electric pump wells with low scores, including:

[0161] The running state of the electric pump well without any fault is scored by the electric pump well operation scoring evaluation system to obtain the electric pump well score;

[0162] For electric pump wells with low electric pump well scores, the number of oilfield electric pump failures is determined through the electric pump well fault evaluation method, and the pump inspection cycles of different types of electric pumps are statistically obtained as the fault evaluation results;

[0163] Based on the fault evaluation results and the static data in the electric pump well related data, the operation effect of the electric pump is evaluated, and the electric pump wells that are of particular concern are obtained;

[0164] Among them, static data include production layer, fluid density, well type, injection-production ratio, pump hanging depth, asphaltene and gas-liquid ratio.

[0165] Among them, the execution process of each unit can be executed according to the process steps of a full-process multi-dimensional production analysis and adjustment method for a carbonate rock electric pump well in Example 1, and will not be repeated one by one in this embodiment.

[0166] At the same time, the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned full-process multi-dimensional production analysis and adjustment method for carbonate rock electric pump wells.

[0167] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0168] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0169] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0170] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0171] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A full-process multi-dimensional production analysis and adjustment method for carbonate rock electric pump wells, characterized in that: The method includes: Acquire the electric pump well related data, form an integrated electric pump well data framework and store it in an electric pump well integrated database; the electric pump well related data includes geological data, reservoir data, wellbore data and electric pump data; Based on the integrated electric pump well data framework, a mechanism diagnosis logic sub-model is constructed; the mechanism diagnosis logic sub-model is an identification logic table composed of preset electric pump well fault types, fault causes and adjustment measures; Based on the integrated electric pump well data framework, a current and voltage diagnosis sub-model, a pump parameter diagnosis sub-model, a pressure and production diagnosis sub-model and a vibration diagnosis sub-model are respectively constructed using a neural network; Using a deep learning neural network, the mechanism diagnosis logic sub-model, the current and voltage diagnosis sub-model, the pump parameter diagnosis sub-model, the pressure and output diagnosis sub-model and the vibration diagnosis sub-model are integrated into a comprehensive diagnosis model; The data of the electric pump well integrated database is input into the comprehensive diagnostic model to determine whether the electric pump well has a fault; if the electric pump well has a fault, the cause of the fault is analyzed based on the fault type of the electric pump well and combined with the identification logic table, and subsequent measures are optimized based on different fault causes and fault types of the electric pump well.

2. A carbonate rock electric pump well full-process multi-dimensional production analysis and adjustment method according to claim 1, characterized in that: The method further includes: If the electric pump well has not failed, the electric pump well that has not failed will be scored through the electric pump well scoring and evaluation system, and attention will be paid to the operating status of the electric pump well with a low score.

3. A full-process multi-dimensional production analysis and adjustment method for carbonate rock electric pump wells according to claim 2, characterized in that: The electric pump well operation scoring evaluation system is used to score the electric pump wells that have not experienced any failures, and attention is paid to the operating status of the electric pump wells with low scores, including: The running state of the electric pump well without any fault is scored by the electric pump well operation scoring evaluation system to obtain the electric pump well score; For electric pump wells with low electric pump well scores, the number of oilfield electric pump failures is determined through the electric pump well fault evaluation method, and the pump inspection cycles of different types of electric pumps are statistically obtained as the fault evaluation results; According to the fault evaluation result and the static data in the electric pump well related data, the operation effect of the electric pump is evaluated to obtain the electric pump well of key concern; The static data include production layer, fluid density, well type, injection-production ratio, pump hanging depth, asphaltene and gas-liquid ratio.

4. A full-process multi-dimensional production analysis and adjustment method for carbonate rock electric pump wells according to claim 3, characterized in that: Evaluate the operation effect of the electric pump, including: The stable operation degree of the electric pump is evaluated as follows: the pump inspection cycle under the screening conditions / the average pump inspection cycle of the entire oil field × 100%. The failure rate of the electric pump is as follows: the number of failures of the electric pump under the screening conditions / the total number of failures of the entire oil field × 100%. The failure rate of the electric pump was evaluated as follows: the number of electric pump failures under the screening conditions / the total number of failures in the entire oil field × 100%.

5. The method for multi-dimensional production analysis and adjustment of the entire process of carbonate rock electric pump wells according to claim 1, characterized in that: The electric pump well fault types include electric pump fault, fluid change fault, unit and pipe string fault; Wherein, the electric pump faults include pump blockage, underload, overload, load fluctuation, overcurrent and undervoltage faults; The fluid changes include fluid impurities and gas content increase faults; The unit and tubing failures include unit aging, undercurrent protection failure, overcurrent protection failure, pump disconnection, downhole tubing leakage and impeller wear failure.

6. The method for multi-dimensional production analysis and adjustment of the entire process of carbonate rock electric pump wells according to claim 1, characterized in that: The steps of constructing the current and voltage diagnosis sub-model include: Batch read the electric pump operation data from the electric pump well integrated database, and perform data preprocessing to obtain time column and current column data; Based on the time column and current column data, coordinate conversion is performed and batch drawing is performed into current card diagrams of standard pixel size; The coordinate conversion is to plot the collected current train data that changes with time into radial coordinates.

7. A carbonate rock electric pump well full-process multi-dimensional production analysis and adjustment method according to claim 6, characterized in that: The conversion expression of the coordinate conversion is: In the formula, x is the horizontal axis value; y is the vertical axis value; t is the time of the day, converted to the hour system; A is the current.

8. A full-process multi-dimensional production analysis and adjustment system for carbonate rock electric pump wells, characterized in that: The system includes: An acquisition unit is used to acquire electric pump well related data, form an integrated electric pump well data framework and store it in an electric pump well integrated database; the electric pump well related data includes geological data, reservoir data, wellbore data and electric pump data; The first sub-model unit is used to construct a mechanism diagnosis logic sub-model based on the integrated electric pump well data framework; the mechanism diagnosis logic sub-model is a recognition logic table composed of preset electric pump well fault types, fault causes and adjustment measures; The second sub-model unit is used to construct a current and voltage diagnosis sub-model, a pump parameter diagnosis sub-model, a pressure and production diagnosis sub-model and a vibration diagnosis sub-model respectively based on the integrated electric pump well data framework by using a neural network; A model fusion unit, used to fuse the mechanism diagnosis logic sub-model, the current and voltage diagnosis sub-model, the pump parameter diagnosis sub-model, the pressure and output diagnosis sub-model and the vibration diagnosis sub-model into a comprehensive diagnosis model by using a deep learning neural network; The fault judgment unit is used to input the data of the electric pump well integrated database into the comprehensive diagnosis model to determine whether the electric pump well has a fault; if the electric pump well has a fault, the cause of the fault is analyzed according to the fault type of the electric pump well and combined with the identification logic table, and subsequent measures are optimized according to different fault causes and fault types of the electric pump well.

9. A full-process multi-dimensional production analysis and adjustment system for carbonate rock electric pump wells according to claim 8, characterized in that: The system also includes: The electric pump operation evaluation unit is used to score the electric pump wells that have not failed through the electric pump well scoring evaluation system if the electric pump well has not failed, and pay attention to the operation status of the electric pump wells with low scores, including: The running state of the electric pump well without any fault is scored by the electric pump well operation scoring evaluation system to obtain the electric pump well score; For electric pump wells with low electric pump well scores, the number of oilfield electric pump failures is determined through the electric pump well fault evaluation method, and the pump inspection cycles of different types of electric pumps are statistically obtained as the fault evaluation results; According to the fault evaluation result and the static data in the electric pump well related data, the operation effect of the electric pump is evaluated to obtain the electric pump well of key concern; The static data include production layer, fluid density, well type, injection-production ratio, pump hanging depth, asphaltene and gas-liquid ratio.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, a full-process multi-dimensional production analysis and adjustment method for a carbonate rock electric pump well is implemented as described in any one of claims 1 to 7.

Images